Classifications

• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
  • H04L49/00—Packet switching elements
  • H04L49/90—Buffering arrangements

Definitions

• the invention relates to the communication of data. More specifically, the invention relates to a technique for controlling data packet transmissions.
• client buffers are commonly used to acquire and selectively deliver packets to the client on an as needed basis based on the content rate (i.e., the rate at which the packets are used to provide content to the client).
• content rate i.e., the rate at which the packets are used to provide content to the client.
• client buffer size is often limited, and a packet transmission rate combined with a link rate that is higher than the content rate will rapidly result in a client buffer overflow. Once overflow has occurred, excess packets are dropped, and with most systems, dropped packets are typically not retransmitted due to bandwidth and delay constraints associated with packet recovery.
• network buffers such as buffers in the SGSN (Serving GPRS Support Node) or in the RNC (Radio Network Controller) within a fixed or mobile network (e.g., IP network) can be utilized to buffer transmitted packets to accommodate varying link rates.
• SGSN Serving GPRS Support Node
• RNC Radio Network Controller
• the link rate because of variations in quality of service (such as an outage of a wireless link resulting from, for example, a handover), dips below the data transmission rate from the server, then packets will begin to accumulate in the network buffer. If these packets exceed the capacity of the network buffer, then packets may be lost which are typically not recovered. If on the other hand, the link rate exceeds the data transmission rate, then the client buffer will receive packets at a rate equal to the data transmission rate.
• Some conventional techniques for controlling packet transmission rates utilize infor- mation within Real Time Transport Control Protocol (RTCP) receiver reports (RRs) to determine whether to adjust the transmission rate (see, for example, Ott et al. "Extended RTP Profile for RTCP-based Feedback", Internet Draft draft-ietf-avt.rtcp- feedback-07.txt (work in progress) June 2003; Schulzrinne et al., "RTP: A transport protocol for real-time applications," RFC 3550, July 2003; and Rey et al., "RTP Re- transmission Payload Format", Internet-Draft draft-ietf-avt-rtp-retransmission-08.txt (work in progress), June 2003).
• RTCP Real Time Transport Control Protocol
• Information pertaining to the oldest buffered sequence number (OBSN) with the client buffer and the highest received sequence number (HRSN) within the client buffer is contained with the RR and is used to determine the consumed buffer space as the size of each packet within the range from the HRSN to the OBSN is known.
• the transmission rate is adjusted based on an estimated rate at which the amount of data within the network buffer is increasing or decreasing relative to a preferred network buffer fill level.
• the transmission rate is adjusted to be equal to the content rate. While this ar- rangement provides certain advantages, the transmission rate increases during link outages in an effort to increase the client buffer level which can result in a network buffer overflow and packet loss.
• the invention may be embodied in a method for controlling packet transmissions from a server coupled via a network buffer (either directly or via a fixed or mobile network such as the Internet) to a client having a client buffer.
• a method includes the steps of determining the network buffer fill level and determining the client buffer fill level (either determination step may first occur or the determination steps may occur simultaneously). If the transmission of the next packet would cause the network buffer fill level to exceed a first threshold and/or the client buffer fill level to exceed a second threshold, then the transmission of the next packet is delayed. These steps may be conducted periodically to ensure that the capacity of both the network buffer and the client buffer does not exceed certain levels (which may be based on their maximum capacities).
• the current invention preferably comprises the step of determining the packet size of the next packet to be transmitted. If the packet size would cause the network buffer or the client buffer to exceed predetermined levels, then the transmission of the packet can be delayed (which indirectly affects the packet transmission rate). Merely adjusting the packet transmission rate, without reference to the size of the next packet (or packets), can result in unwanted buffer overflow as the rate at which the packet is transmitted is irrelevant if there is not sufficient capacity within the buffers to buffer the packet.
• Determination of the network buffer fill level may be based on the aggregate size of packets within the network buffer.
• the network buffer may comprise a plurality of individual buffers that may at least partially be coupled in series, and so a determination must be made as to the fill level of one or more (or each) of these buffers to ensure that the transmission of the next packet would not cause a buffer overflow.
• the first threshold may be based, for example, on the buffer having the lowest capacity or the capacity of the buffer nearest to the wireless transmitter, or alternatively, each of the buffers may be individually monitored and individual thresholds may be assigned.
• the client buffer fill level may be determined based on the aggregate size of packets within the client buffer.
• the client buffer fill level determination may also be based on the aggregate size of packets within the client buffer and the aggregate size of packets within the network buffer. The latter arrangement assumes a scenario where all of the packets within the network buffer are immediately transmitted to the client buffer and can be used in arrangements where it is important to avoid client buffer overflow.
• the client buffer fill level may be determined based on the aggregate size of packets within the client buffer and the aggregate size of packets within selected ones or within all of the individual buffers that constitute the network buffer. In such an arrangement, the con- tent of one or more buffers closest to the client buffer (in the transmission direction) may be aggregated.
• the buffer fill level determinations may be aided by assigning each of the packets with sequential identification numbers (or codes). Such an arrangement may permit the determination of the network buffer fill level to be based on an aggregate sum of the size of packets ranging between the highest sequence number packet transmitted by the server and the packet having a sequence number immediately preceding or equaling the highest sequence number of a packet within the client buffer.
• the client buffer fill level in some embodiments, may either be determined based on an aggregate sum of the size of packets ranging between the highest sequence number packet and the lowest sequence number packet within the client buffer, or it may be determined based on an aggregate sum of the size of packets ranging between the highest sequence number packet transmitted by the server and the lowest sequence number packet within the client buffer.
• Information pertaining to the packets within the network buffer and the client buffer may be obtained through a variety of mechanisms, such as via periodic feedback reports from the client.
• the periodic feedback reports might contain information as- sociated with at least one of client buffer fill level status, highest received sequence number by the client buffer, and oldest buffered sequence number.
• One type of feedback report that might be used with the current invention is a Real-Time Transport Control Protocol receiver report.
• the determination regarding the network buffer fill level and/or the client buffer fill level may be based on an estimate of the amount of time for a transmitted packet to be played by the client after transmission by the server.
• the first threshold which, according to the invention, may be set as a maximum or target fill level for the network buffer, may be set based on the maximum network buffer capacity.
• the second threshold may be set based on the client buffer maximum capacity. Both of the thresholds may be based on a percentage of the maximum buffer capacities, such as 95 percent, to further ensure that neither buffer is subject to overflow conditions.
• the first threshold may also be determined by the steps of assigning an initial threshold value, monitoring whether the network buffer overflows, increasing the threshold value from the initial threshold value if a network buffer overflow does not occur within a predefined period of time, and/or reducing the threshold value from the initial threshold value if a network buffer overflow occurs. This process may be repeated a predetermined number of times or for a predetermined amount of time, provided that the resulting threshold value is not too high to cause network buffer overflow and too low to increase the likelihood of client buffer underflow.
• the method may also include the step of determining the maximum number of packets that may be buffered by at least one of the network buffer and the client buffer. This determination may be based on information pertaining to overflows in the buffers or it may be obtained from pre-existing measurements. These determinations may then be used to set the values for the first and second thresholds.
• the method in some variations, includes the step of (indirectly) adjusting the transmission rate.
• the transmission rate is adjusted to approximately maintain the network buffer at approximately the first threshold or the client buffer at the second threshold.
• the method may act to periodi- cally adjust the packet transmission rate to the maximum level possible while avoiding an overflow of either of the buffers.
• the client may be a mobile communications terminal such as a mobile telephone
• the server may be integrated into a mobile communications terminal so that the link between the server and the network may be wireless.
• the method according to the present invention may be performed by one or more intermediary network nodes (such as proxies) arranged between the server and the client.
• the method may also be utilized in architectures having plurality of data streams being buffered by the network buffer and/or client buffer (or multiple client buffers depending on the configuration of the client).
• the invention may also be embodied in a computer program product, which may be stored on a computer readable recording medium, comprising program code portions for performing any of the steps of the above methods when the computer program product is run on a computer system.
• the invention may further comprise an apparatus comprising a computer processor and a memory coupled to the processor, where the memory is encoded with one or more programs that may perform any of the steps of the above methods.
• the invention in yet another embodiment, relates to an apparatus for controlling packet transmissions from a server via a network buffer to a client having a client buffer.
• the apparatus comprises a first determination unit for determining the network buffer fill level, a second determination unit for determining the client buffer fill level, and a control unit for delaying transmission of the next packet if the packet would cause either the network buffer fill level to exceed a first threshold or the client buffer fill level to exceed a second threshold.
• the apparatus may be configured as a fixed or mobile network component, such as a network server and/or a wireless terminal. Also, the apparatus may be constituted by an intermediary network node, such as a proxy.
• Fig. 1 is a first schematic diagram useful for understanding and implementing the invention
• Fig. 2 is a second schematic diagram of useful for understanding and implementing the invention
• Fig. 3 is a process flow diagram of method embodiment of the invention
• Fig. 4 is a process flow diagram of an example useful for understanding and implementing the invention.
• Fig. 5 is a schematic of an apparatus embodiment of the invention:
• Fig. 6A is a first chart graphically illustrating a network buffer fill level relative to a network buffer maximum capacity for a first sample data run using a first conventional technique
• Fig. 6B is a second chart graphically illustrating a network buffer fill level relative to a network buffer maximum capacity for the first sample data run using a second conventional technique
• Fig. 6C is a third chart graphically illustrating a network buffer fill level relative to a network buffer maximum capacity for the first sample data run using the technique provided by the current invention
• Fig. 7A is a first chart graphically illustrating a client buffer fill level relative to a client buffer maximum capacity for the first sample data run using a first conventional technique
• Fig. 7B is a second chart graphically illustrating a client buffer fill level relative to a client buffer maximum capacity for the first sample data run using a second conventional technique.
• Fig. 7C is a third chart graphically illustrating a client buffer fill level relative to a client buffer maximum capacity for the first sample data run using the technique provided by the current invention.
• Fig. 1 illustrates a sample architecture 100 that may be used in connection with the invention including a server 105 that may be coupled to a client 115 via a communications pathway such as an IP network 110.
• the server include a media content module 120 that accesses and transmits certain media content (e.g., multimedia data files) via a RTP/UTP module 125 using a streaming standard such as RTP (Real time Transport Protocol) over UTP or other data transport protocol for managing the realtime transmission of multimedia data (with a transport layer protocol such as UDP - User Datagram Protocol).
• RTP Real time Transport Protocol
• the packets are transmitted to a public network 130 (e.g., the Internet, however, an external public network is not required when the server is directly coupled to the operator network 135) that delivers the packets to an operator network 135, such as a mobile communications operator "wired" network, coupled thereto.
• a public network 130 e.g., the Internet, however, an external public network is not required when the server is directly coupled to the operator network 135
• an operator network 135 such as a mobile communications operator "wired" network, coupled thereto.
• the operator network 135 includes a core network 140 that provides communication links between the server 105 and the client 115.
• the core network 140 which may optionally have a buffer, provides the packets received from the RTP/UTP module 125 for buffering in a buffer within a radio access network (RAN) 145 (such as a buffer in the SGSN or in the RNC) prior to their transmission by a wireless transmitter 150.
• RAN radio access network
• the buffers of the core network 140 (if buffering is utilized) and the RAN 145 are arranged in series and constitute a network buffer.
• the client 115 receives the packets transmitted by the wireless transmitter 150 in a client buffer 155.
• the packets are transferred from the client buffer 155 to a RTP/UTP module 160 for delivery to and use by the media application module 165 (or multimedia player).
• the phrase "packet transmission rate” will refer to the rate of transmission of packets from the server 105 to the IP network 110
• the phrase “link rate” will refer to the rate of transmission of packets from the IP network 110 to the client 115
• the phrase “content rate” shall refer to the rate that data is transferred from the client buffer 115 to the media application module 165 for playout by the media application module 165.
• Fig. 2 relates to a particular architecture 200 (although it will be appreciated that certain of the components may be combined and implemented) that may be used with the current invention having a server 205 that communicates with a client 215 via an IP network 210. Packets that are to be transmitted are received by the server 205 via an input stream module 220 that delivers the packets to an RTP module 225 for processing and transmission over the IP network 210 and are buffered by a network buffer 230 until they are transmitted to the client 215. The transmitted packets are received by the client 215 via an RTP module 235 which delivers the processed packets to a client buffer 240.
• the client 215 further includes an RTCP module 250 coupled to the client buffer 240 for providing feedback (e.g., an RR) to an RTCP module 255 within the server 205.
• the RTCP information contains quality of service feedback including, inter alia, the HRSN and OBSN within the client buffer 240 as well as Playout Delay (PD).
• the server 205 evaluates the RTCP information using a feedback evaluation module 260. Based on this evaluation, the server 205 may elect to adjust the transmission rate of packets received by the input stream module 220 via a transmission rate control module 265. This transmission adjustment may include the complete cessation of data transmission (i.e., no further packets will be transmitted by the server 205 to the client 215), or it may include an upward or downward adjustment of the data transmission rate depending on criteria described below.
• a method embodiment 300 of the invention is illustrated which may be used, for example, with a system such as those illustrated in Figs. 1, 2 and 5.
• the method begins, at step 310, with the determination of the network buffer fill level, or in other words, the aggregate size of accumulated packets contained within the network buffer.
• the method also determines, at step 320, the client buffer fill level, or in other words, in some cases, the aggregate size of accumulated packets contained within the client buffer.
• the transmission of pack- ets are delayed by the server, if the network buffer fill level would exceed a first threshold and / or if the client buffer fill level would exceed a second threshold, if the next packet is transmitted.
• the next packet or packets will be transmitted by the server (which preferably is at a maximum transmission rate that complies with both of these conditions). Examples of how to determine the network and client buffer fill levels and the first and second thresholds are described below.
• the server assigns a sequence number to each transmitted packet.
• the server may store the sequence numbers of transmitted packets as well as the size of each packets in a server database (which may be internal or external to the server).
• the server database may be accessed to provide information such as the Highest Transmitted Sequence Number (HTSN) corresponding to the sequence number of the last transmitted packet and the size of each transmitted packet.
• HTSN Highest Transmitted Sequence Number
• This information may be used to determine the network buffer fill level, which may be based, in part, on the cumulative size of all transmitted packets up to sequence number j, A Jrl which may be calculated using the following formula, where Si is the size in bytes of packet with sequence number /
• the network buffer fill level, ⁇ is equal to the cumulative size of all packets which have not been received by the client buffer according to the last RR, but which have actually been transmitted by the server. This value is calculated as:
• Ntev AHTSN - AHRSN
• f ⁇ ? is the time at which the RR arrives at the server.
• the server can estimate the last played sequence number (LPSN) as follows:
• the client buffer fill level Q ev may be estimated using a formula such as:
• a "worst-case scenario" client buffer occupancy may be calculated that assumes that all packets currently buffered in the network buffer are transferred to the client buffer.
• the client buffer fill level is calculated as follows (where HRSN has been substituted with HTSN):
• the server may set maximum fill levels for both of the network and client buffers, ⁇ and C ma x respectively.
• C max ⁇ s set to a value slightly smaller than the actual client buffer size (which can be signalled to the server via RTSP during session initialization), and /I is based on pre-existing information regarding the IP network buffer size or, in some cases, the network buffer fill level may be estimated (and if the network buffer comprises multiple buffers, ⁇ may be based on the buffer having the lowest capacity).
• N max may be set by assigning a first threshold value corresponding to an estimated network buffer maximum capacity (which may be based on factors such as average network buffer size, expected network buffer size for IP network, etc.). If the network buffer overflows, then ⁇ is incrementally lowered until a condition exists where the network buffer does not periodically overflow.
• the server will not transmit additional packets unless both of the following conditions are met:
• a packet is selected for transmission. It is then determined, at step 420, whether the packet will fit into the network buffer, or in other words, whether the addition of the selected packet will be equal to or greater than Nmax. If the addition of the selected packet into the network buffer would not exceed the maximum fill level, then it is determined, at step 430, whether the selected packet, and optionally the packets within the network buffer, would fit into the client buffer (which may already have packets being buffered therein). If this condition is met, then at step 440, the server sends the selected packet. If the criteria of either step 420 or 430 is not met, then step 420 is repeated (at periodic intervals) until the next packet may be sent.
• a network component 500 useful for practicing aspects of the invention in connection with a server, a network having a network buffer, and a client having a client buffer, is illustrated in Fig. 5.
• the network component 500 (which may be a network node such as a proxy, a wireless terminal, and the like) includes a first determination unit 520 which determines the network buffer fill level, a second determination unit 530 that determines the client buffer fill level, and a control unit 510 coupled to the first determination unit 520 and the second determination unit 530.
• control unit 510 determines that either the transmission of the next subsequent packet within a transmission queue by the server would cause either of the network buffer or the client buffer to respectively exceed first and second capacity thresholds, then the control unit 510 will send a message to a transmitting server to temporarily cease the transmission of packets. If the transmission of the next packet will not cause the capacity of the network buffer to exceed the first threshold and/or the client buffer to exceed the second threshold, then the control unit 510 will either send a new message to the server to recommence packet transmissions (in the case that a message was previously sent to cease transmission), or the control unit 510 will not send a cessation signal to the server (i.e., a signal will not be sent if there is no action that needs to be taken by the server).
• Figs. 6A - 6C demonstrate the improvements of the current invention with regard to network buffer fill level as compared to conventional techniques.
• Fig. 6A in graph 600, illustrates the network buffer fill level (in bytes) as compared to the network buffer maximum capacity (with one incident of network buffer overflow) for a simulated data transfer using a Constant Rate Transmission (CRT) technique.
• Fig. 6B in graph 610, illustrates the network buffer fill level as compared to the network buffer maximum capacity (with one incident of network buffer overflow) for a simulated data transfer using a Proportional Derivative Feedback Control (PDFC) rate technique.
• PDFC Proportional Derivative Feedback Control
• Fig. 6C in graph 620, illustrates the network buffer fill level as compared to the network buffer maximum capacity (without network buffer overflow) for a simulated data transfer using a technique according to the current invention.
• Figs. 7A - 7C demonstrate the improvements of the current invention with regard to client buffer fill level as compared to conventional techniques.
• Fig. 7A in graph 700, illustrates the client buffer fill level as compared to the client buffer maximum capacity (with one incident of client buffer underflow) for a simulated data transfer using the CRT technique.
• Fig. 7B in graph 710, illustrates the client buffer fill level as compared to the client buffer maximum capacity (with one incident of client buffer underflow) for a simulated data transfer using the PDFC rate technique.
• Fig. 7C in graph 720, illustrates the client buffer fill level as compared to the client buffer maximum capacity (without client buffer underflow) for a simulated data transfer using a technique according to the current invention.
• the link rate was the bottleneck and that the bandwidth of the wired network (i.e., the network between the server and the IP network) was not a limiting factor (such as when the server is either located directly on the operator network or connected to it by a non- congested public network).
• the simulated data transfer used a wireless link having a 64 kbps bandwidth with a five second complete link outage between seconds 18 and 23.
• a H.263 VBR video sequence consisting of 359 packets with an average bit rate of 57 kbps was used).
• a five second pre-buffering time was used for the playout by the media application (e.g., multimedia player).
• a 50 KB client buffer and a 20 KB network buffer were also assumed. Apart from losses due to overflow in these two buffers, no other losses were assumed during transmission, including in the wireless link where the existence of a link-level ARQ mechanism providing reliable transmission was assumed.
• RTCP Sender and Receiver Reports were imple- mented, with a RR interval of one second. NACK reports were issued and retransmissions, when estimated RTT showed arrival at the receiver before presentation time to be likely, were carried out with a conventional Early Deadline First strategy.
• the CRT technique used a transmission rate 60 Kbps and is representative of the generic category of conventional piece-wise constant rate transmission techniques because the transmission rate is chosen in order to provide satisfactory client buffer management in cases where available bandwidth is not a concern.
• Figs. 6A and 7A as CRT is not concerned in maximizing client buffer fill level, and therefore a client buffer underflow with consequent rebuffering occurs shortly after the link outage. Moreover, as a consequence of the rebuffering event, the client buffer tends to underflow.
• the PDFC technique used RTCP feedback information may be represented by the following equation:
• n indicates the instant at which each RR is received
• R is the transmission rate
• C t is the target level for client buffer occupancy
• ⁇ t ⁇ n _ is the time interval between current and last RR
• K and K 2 are the feedback gains of the proportional and the derivative term respectively.
• the current invention provides the optimum network and client buffer fill levels, without network buffer overflows and client buffer underflows, compared to the conventional techniques, thereby avoiding packet loss and performance degradation associated therewith.
• Table 1 below lists the number of packets lost using the respective techniques as well as the number of packets that were not played out as a result of these overflows for the simulated data transfer.
• the current invention and the techniques associated therewith provide an enhanced end user perceived experience for applications such as multimedia streaming by avoiding both network and client buffer overflows.
• the invention may be used standalone or it may be used in connection with packet scheduling or content rate control algorithms which adapt the media content rate based on the link rate.
• the invention may be implemented in and by various types of network components, such as network terminals, network nodes, and the like.
• the invention may be practiced by mobile terminals, proxies (that could divide the transmission path), and fixed terminals.

Landscapes

• Engineering & Computer Science (AREA)
• Computer Networks & Wireless Communication (AREA)
• Signal Processing (AREA)
• Data Exchanges In Wide-Area Networks (AREA)
• Communication Control (AREA)
• Auxiliary Devices For And Details Of Packaging Control (AREA)

Applications Claiming Priority (1)

Publications (2)

Family

ID=34878401

Family Applications (1)

Country Status (6)

Families Citing this family (7)

Family Cites Families (4)

• 2004
  • 2004-02-20 EP EP04713055A patent/EP1716672B1/de not_active Expired - Lifetime
  • 2004-02-20 DK DK04713055T patent/DK1716672T3/da active
  • 2004-02-20 AT AT04713055T patent/ATE427607T1/de not_active IP Right Cessation
  • 2004-02-20 WO PCT/EP2004/001693 patent/WO2005081465A1/en not_active Application Discontinuation
  • 2004-02-20 DE DE602004020391T patent/DE602004020391D1/de not_active Expired - Lifetime
• 2005
  • 2005-01-28 TW TW094102707A patent/TWI333359B/zh active

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events